Whilst it is common in clinical trials to use the results of tests at one phase to decide whether to continue to the next phase and to subsequently design the next phase, we show that this can lead to biased results in evidence synthesis. Two new kinds of bias... ...
